ByteDance-Seed / ByteDance-Seed/decoupleQ

量化结果差异咨询

Open
#19 1 comment 0 reactions 0 assignees View on GitHub
Dominant language
Cuda
Stars
153
Forks
10
PR merge metrics
No merged PRs in 30d

Description

作者你好,基于llama 2-7B模型
wikitext-2:量化前:5.47(和论文一致),**W2A16量化后10.22(论文写的9.74)**。W4A16 量化后5.70(和论文一致)
想请教一下,W2A16量化后的差异可能是因为什么原因导致的呢?我所有的参数都是采用默认值~

Contributor guide

No contributing guide indexed for this repository

Research direction

The report compares Llama 2-7B W2A16 and W4A16 quantization results on wikitext-2 using default parameters, but names no files, tests, or entry points. Start by reproducing the reported W2A16 result with the repository's default workflow and compare it with the paper's evaluation setup. Done means identifying and documenting the cause of the 10.22 versus 9.74 discrepancy.

Written by the indexing model from the issue text.

Assessment

Domain
machine-learning
Issue type
Bug
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
20/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.